It seems relevant that the categories healthcare, housing, and higher education are not paid by wages/cash but rather through various means of financing (in the form of insurance for healthcare). (In addition to not being outsourcable).

New cars not increasing looks strange but that might be a US phenomena, where most cars might be imported and the median car have "shrunk" in size over the period.

It highly depends on what your definition of a price increase is. The cars people actually buy have definitely become more expensive with more and more people choosing to buy SUVs.

The sedan hasn't become more expensive, but people don't feel safe driving them anymore between all the large SUVs, this pushing more people towards buying more expensive cars.

> MDF and veneers are inherently worse than hand-crafted wood.

Generally incorrect, but it depends. Wear can cause mdf/veneer to have "bad optics" compared to solid wood, but mdf/veneer can have more suitable physical properties and enables more consistent visual quality and design possibilities.

I suppose it depends on your definition of worse. It is more versatile. It's also toxic and fragile, and far more likely to break in ways that are hard to repair. I can only think of one object I own where the physical properties of particle board or MDF are a positive: a subwoofer where its consistency helps with acoustics.

It helps with all types of speakers, not just subwoofers. For the same reasons, it can work as structural parts of furniture.

A 2015 smartphone could be eight 2 GHz cores with 2-4 GB memory and 128 GB SSD. By far faster and more powerful than anything but the heaviest 2000-era servers if that.

A hidden markov model (HMM) is theoretically capable of modelling text just as well any transformer. Typically, HMMs are probability distributions over a hidden discrete state space, but the distribution and state space can be anything. The size of the state space and transition function determines its capacity. RNNs are effectively HMMs, and recent ones like "Mamba" and so on are considered competent.

Transformers can be interpreted as tricks that recreate the state as a function of the context window.

I don't recall reading about attempts to train very large discrete (million states) HMMs on modern text tokens.

"had managed to increase revenues by 108%"

This sounds less than the consumer price inflation over the period.

Average of costs of compensation - "including employee-related taxes and pension contributions".

The employee sees around half of that. But there must be more to it, since it is still an unrealistic number.

"100.000 servers" sounds like the entirety of 2008-era Google, but with today's 10-100x performance per server. It could be bigger than Dropbox or Netflix.

No, it supposedly makes things more efficient and thus cheaper.

The number of iphones sold per year is about constant over the last decade. The amount of materials and work-hours in each iphone is probably constant. So if there are no alternatives, "number go up".

The phenomenon is probably a consequent of monetary inflation and lock-in.

Yep this is what I thought as well. All this is probably linked to money oversupply and generalized inflation. It is particularly noticeable on the non-essential items, like tech stuff.

And revenue is a weird indicator to take; profit would make more sense. We know tech gets very good effects of scale but it seems there are limits even for software.